Mitochondrial Health Index Correlates with Plasma Circulating Cell-Free Mitochondrial DNA in Bipolar Disorder

Background: Although mitochondria dysfunction is known to play an essential role in the pathophysiology of bipolar disorder (BD), there is a glaring gap in our understanding of how mitochondrial dysfunction can modulate clinical phenotypes. This study aimed to evaluate the composite mitochondrial health index (MHI) in BD subjects and non-psychiatry controls (Non-psychiatry controls). We will also explore whether lower MIH will be related to higher cell-free mtDNA (ccf-mtDNA) levels and poor clinical outcomes. Methods: Fourteen BD-I patients and 16 age- and sex-matched non-psychiatry controls were enrolled for this study. Peripheral blood mononuclear cells (PBMCs) were used to measure the enzymatic activities of citrate synthase and complexes I, II, and IV and mtDNA copy number. ccf-mtDNA was evaluated by qPCR in plasma. Mitochondrial quality control (MQC) proteins were evaluated by western blotting. Results: One-Way ANCOVA after controlling for age, sex, body mass index (BMI), and smoking status showed that patients with BD present a decrease in the MHI compared to non-psychiatry controls, and higher ccf-mtDNA levels, which was negatively correlated with MHI. Because the MQC network is essential to maintain mitochondrial health, we also evaluated the relationship between MQC-related proteins with MHI and ccf-mtDNA. Our results showed that MHI negatively correlated with Fis-1 and positively with Opa-1 and LC3. Moreover, we found a negative correlation between ccf-mtDNA, Opa-1, and LC3 and a positive correlation between cff-mtDNA and Fis-1. Finally, we found that subjects with longer illness duration, higher depressive symptom scores, and worse functional status had lower MHI and higher ccf-mtDNA. Conclusion: In summary, the present findings corroborate previous studies and provide strong support for the hypothesis that mitochondrial regulation and function are integral parts of the pathogenesis of BD.


Introduction
Bipolar disorder (BD) is a severe and chronic psychiatric disorder that affects approximately 1-4% of the world population 1 . BD seems to be a steadily growing threat to the health systems since current treatment options remain unsatisfactory, and many patients continue to experience intra-episodic symptoms or even full-blown episodes resistant to treatment 2 . The pathophysiological pathways responsible for BD remain elusive even after many years of research, likely due to multifactorial etiology involving the interaction between multiple genetic, neurochemical, and environmental factors 3 .
The mitochondrial dysfunction hypothesis has been corroborated by several studies showing that BD patients present an atypical mitochondrial metabolism, oxidative stress, abnormal mitochondrial morphology and dynamics, and mitochondrial DNA (mtDNA) damage 4 . Genetic studies also supported this hypothesis, showing an increased likelihood of maternal inheritance in generational transmission of BD, abnormal ndings in the mtDNA of patients with BD, and comorbidity of affective disorder with mitochondrial diseases [5][6][7] . Moreover, chronic stress (e.g., recurrent affective episodes) can lead to an accumulation of mitochondrial dysfunction -or mitochondrial allostatic load -which might represent an early event that increases allostatic load and disease risk in BD 8, 9 . To disentangle these contributors and identify the molecular nature of a potential mitochondrial perturbation, a mitochondrial health index (MHI) was developed to assess mitochondrial functional capacity in human leukocytes 10 . This metric integrates nuclear and mitochondrial DNA-encoded respiratory chain enzymatic activities and mtDNA copy number (mtDNAcn) into an index re ecting mitochondrial respiratory chain (RC) capacity per-mitochondrion basis. The MHI was previously found to be low among highly stressed caregivers compared to controls, and the MHI was also associated with mood states in this group 10 . While these ndings are novel and interesting, the MHI has not, to date, been applied to BD, and its clinical relevance remains unknown.
Although preliminary studies from our group suggest that BD patients present an impairment in mitochondrial dynamics and mitophagy, it is unknown whether the failure to clear damaged mitochondria mediates the decrease in mitochondrial function. We hypothesize that BD patients will show lower MHI due to changes in the MQC network, which will be associated with higher cell-free mtDNA (ccf-mtDNA) levels and poor clinical outcomes. To test this hypothesis, in this preliminary study, we evaluate the activities of mitochondrial complexes I, II, IV, citrate synthase (CS), and mtDNAcn, to further calculate the composite MHI in peripheral blood mononuclear cells (PBMCs) from patients with BD and nonpsychiatric controls. We also evaluate the levels of proteins involved in mitochondrial dynamics, mitophagy, active caspase-3, and ccf-mtDNA levels.

Subjects
This study was carried out in accordance with the principles of the Declaration of Helsinki with approval from the Institutional Review Board of the University of Texas Health Science Center at Houston (HSC-MS-09-0340), and written informed consent was obtained from all research participants. Fourteen participants with BD type I (BD-I) were recruited from the UTHealth Mood Disorders outpatient clinic and sixteen non-psychiatry controls were recruited from the local community who did not have a personal psychiatric disorder or family history of major psychiatry disorder or neurologic disorders in rst-degree relatives. Patients and non-psychiatry controls were matched for ethnicity, age, and sex. Patients and nonpsychiatry controls were assessed with the Mini-International Neuropsychiatric Interview (MINI) to con rm BD diagnosis (patients) or to exclude a history of psychiatric disorders (controls) 11 . BD participants were also assessed using the Montgomery-Åsberg Depression Rating Scale (MADRS) 12 and the Young Mania Rating Scale (YMRS) 13 to index the severity of depressive and manic symptoms, respectively. Functioning was assessed with the Global Assessment of Functioning (GAF) Scale and Functioning Assessment Short Test (FAST) 14, 15 . Processing Whole Blood Samples Human blood samples were collected in heparin-coated collection tubes. Then, PBMCs were separated using LeucoPREP brand cell separation tubes (Becton Dickinson Labware, Lincoln Park, NJ, USA). PBMC cell pellets were mixed with RPMI-1640 medium containing 10% DMSO and frozen overnight in a Mr.
Frosty container with 2-propanol (#5100-0001, Nalgene, Rochester, NY) at − 80°C following an appropriate post-processing delay. Preanalytical characterization and quality control (QC) were performed using trypan blue-based methods to evaluate cell viability after (post-thaw) cryopreservation. Our results showed that cell viability remained above 70% in all samples.

Mitochondrial Enzymatic Activities
PBMC pellets were mechanically homogenized in a homogenization buffer extraction buffer containing 1mM EDTA and 50mM triethanolamine to release individual enzymes. Then the homogenate was used to quantify the enzymatic activities of enzymatic activities citrate synthase, complex I, complex II, and cytochrome c oxidase (complex IV) using kinetic spectrophotometric assays. On the day of the assays, the samples were frozen and thawed in hypotonic assay buffer three times to expose the enzymes to substrates and achieve maximal activities fully. Citrate synthase activity was assayed according to the method described by Srere 16 , measuring the formation of the -SH group released from CoA-SH using the reactive Ellman reagent (5,5`-dithiobis [2-nitrobenzoic], DTNB) and monitoring the absorbance at 412 nm. NADH dehydrogenase (complex I) was evaluated according to the method described by Cassina and Radi 17 by the rate of NADH-dependent ferricyanide reduction at 420 nm. The activity of succinate: 2,6dichlorophenolindophenol (DCIP) oxidoreductase (complex II) was determined according to the method of Fischer and colleagues 18 . Complex II activity was measured following the decrease in absorbance due to the reduction of 2,6-DCIP at 600 nm. The activity of cytochrome c oxidase (complex IV) was assayed according to the method described by Rustin and colleagues 19 , measured by following the decrease in absorbance due to the oxidation of previously reduced cytochrome c at 550 nm. The speci c activity for each enzyme was obtained by calculating the slope ( rst derivative) of the optical density change and subtracting non-speci c activity detected in the presence of speci c inhibitors for each complex or in the absence of the rate-limiting reaction substrate. All assays were performed in triplicates at 30 o C, and nal values were normalized on a per-cell basis using the qPCR-based estimates of cell numbers for each biological sample as described in Picard et al 10 .

Mtdna Copy Number
Real-time quantitative PCRs were performed to measure the amount of mtDNA relative to a single-copy gene (beta-hemoglobin) with a modi ed protocol from Tyrka et al 20 . Reactions included 25 ng genomic DNA, 300 nmol l − 1 of each primer, and 1 × Sybr Select Master Mix (Life Technologies, Carlsbad, CA, USA) in a nal volume of 10 µl. Primer sequences and PCR cycling conditions for both mtDNA and beta-hemoglobin have been previously reported 20 . Reactions were carried out in 96-well plates, and data were acquired in a QuantStudio 7 Flex Real-Time PCR System (Life Technologies). mtDNAcn for each sample was determined by relative quanti cation based on a 5-point standard curve performed with a serial dilution (1:2) of a calibrator sample ranging from 1 to 0.0625 ng DNA. All samples were analyzed in triplicate. The relative amount of mtDNA was nally divided by the relative amount of the betahemoglobin gene to obtain an index of mtDNAcn.

Rationale For Calculating The Mhi
The MHI was computed by integrating three enzymatic measures of respiratory chain capacity and two mitochondrial content features, as described previously 10,21 . To calculate the MHI, the four mitochondrial features were mean-centered, so each parameter would contribute an equal weight in the equation. Combining three features (complexes I, II, and IV) as a numerator, divided by two content features (CS and mtDNAcn) as a denominator produces a quantitative index of mitochondrial energy production capacity, or "quality" on a per cell mitochondrion basis, where a value of 100 represents the average of the cohort, and values > 100 and < 100, respectively, indicate higher and lower respiratory chain capacity on a per mitochondrion basis. Previously, the composite MHI exhibited a higher predictive ability of caregiver (chronic stress) status than any of the individual MHI components alone 10,21 .

Plasma Circulating Cell-free Mitochondrial Dna (Ccf-mtdna)
DNA was isolated from thawed plasma samples from the same subjects included in the study using the QIAmp 96 DNA Blood Kit (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. The quantitative analysis of ccf-mtDNA was performed using real-time PCR using SYBR Green Technology

Statistical analysis
Statistical analyses were performed using Statistical Package for the Social Sciences, v.23.0 (SPSS Inc., USA). The normality of data distribution was assessed using the Shapiro-Wilk test and histogram visualization. The subjects' demographic and clinical characteristics were presented in tables as percentages, mean (SD), or median (interquartile range), according to distribution data. Chi-squared was applied for statistical comparisons between the categorical variables. Age, body mass index (BMI), MADRS, YMRS, FAST, and GAF showed nonparametric distributions and were therefore analyzed by the Mann-Whitney U test.
To ensure that linear regression models met their assumptions, regression diagnostics were performed.
Shapiro-Wilk tests were used to determine whether the data distributions were normal. For variables that did not follow a normal distribution, natural logarithms were applied to ensure normality. In addition, standardized residual plots versus standardized predicted values were used to detect linearity and homoscedasticity. Each model was tested for multicollinearity using tolerance criteria and variance in ation factors (VIFs). Diagnostic tests were passed by all models.
Analysis of covariance (ANCOVA) was performed to determine the relationship between groups with the addition of age, sex, BMI, and self-report smoking status (yes/no) to check for signi cant effects and interactions since these factors are thought to interfere with mitochondrial function [23][24][25] . If an interaction was signi cant, Bonferroni corrected post-hoc test was performed (P < 0.05, two-tailed).
Correlations between variables were tested with either Pearson's or Spearman's tests depending on their distribution. All statistical tests were two-tailed and p values < 0.05 were considered statistically signi cant after Bonferroni correction for multiple testing. Linear regression was performed with MADRS, YMRS, GFA, FAST, MHI and ccf-mtDNA as the dependent variables, and MHI, ccf-mtDNA, length of illness as predictor, controlling for sex, age, BMI and self-report smoking status (yes/no).

Results
Characteristics of BD participants and non-psychiatry controls are shown in Table 1. Notably, no signi cant differences in sociodemographic factors (ethnicity, age, sex, BMI, and smoking status) emerged between groups. The mean YMRS score of the BD group was 7.93 ± 7.51, and their mean MADRS score was 17.43 ± 10.61. Moreover, a signi cant difference in functional status, assessed by GAF (p < 0.001) and FAST (p < 0.001), was found between non-psychiatry controls and patients with BD. All patients were on treatment with various psychiatric medications at conventional doses at the time of the study, including antipsychotics, anticonvulsants, mood stabilizers, and antidepressants. Some patients received additional benzodiazepines and stimulants. One-Way ANCOVA, after controlling for age, sex, BMI, and smoking status, showed that patients with BD present lower enzymatic activity of the citrate synthase and complexes I, II, and IV when compared to Non-psychiatry controls. However, the mtDNAcn yielded no statistically signi cant group differences ( Table 2). The present study is the rst to demonstrate that patients with BD had a lower MHI than nonpsychiatry controls after adjusting for confounding variables (Fig. 1). We have previously identi ed an impairment of the MQC in BD patients, described by an imbalance in mitochondrial dynamics towards ssion and reduced levels of proteins responsible for removing damaged mitochondria via mitophagy, followed by apoptosis activation 26, 27 . In the present study, as previously shown by our group, BD patients presented lower levels of Mfn-2, Opa-1, Parkin, p62/SQSTM1, and LC3, while the levels of Fis-1 and active caspase-3 were higher when compared to non-psychiatry controls (Sup. Table 1). Because the MQC system is essential to maintain mitochondrial health, we evaluated the relationship between MQC-related proteins and MHI. Our results showed that subjects with higher levels of Fis-1 exhibited lower MHI (rho = -0.579, p = 0.001), while subjects with higher levels of Opa-1 (rho = 0.616, p < 0.001) and LC3 (rho = 0.600, p = 0.001) had higher MHI (Sup. Table 2).
Since an imbalance between the MQC network and apoptosis activation can lead to the release of mtDNA into the circulation, we evaluated ccf-mtDNA and whether MQC-related proteins, active caspase-3, and MHI could be associated with ccf-mtDNA. Our preliminary data showed that BD patients had higher levels of ccf-mtDNA ( Fig. 2A), which was negatively correlated with MHI (Fig. 2B). In order to investigate whether MHI predicted ccf-mtDNA, we performed linear logistic regression models including age, sex, BMI, and smoking status as covariates. Linear regression analysis showed a negative signi cant predictive relationship between MHI and ccf-mtDNA after controlling for age, sex, BMI and smoking status (R 2 adj = 0.319, F(5, 23) = 3.629, p = 0.014, β = -3.601 p < 0.001). Moreover, our preliminary analyses showed that subjects with higher levels of ccf-mtDNA had higher levels of Fis-1 (rho = 0.538, p = 0.003) and lower levels of Opa-1 (rho = -0.441, p = 0.017) and LC3 (rho = -0.501, p = 0.008) (Sup. Table 2). Additionally, we found that patients with higher levels of active caspase-3 had lower MHI (rho = -0.512, p = 0.005) and higher ccf-mtDNA levels (rho = 0.637, p < 0.001) (Sup. Table 2).
In order to investigate whether MHI and ccf-mtDNA levels predicted mood symptoms and functional impairment we performed linear regression models relating MHI and ccf-mtDNA to one measure of mood (MADRS and YMRS), functional status (GAF and FAST), or clinical status (length of illness), controlling for age, sex, BMI and self-reported smoking status. As can be seen in Table 3, within individuals with BD, MHI and ccf-mtDNA predicted the severity of depressive symptoms, while no association was found with YMRS score. Moreover, we observed a signi cantly negative predictive relationship between length of illness and MHI, and a positive predictive relationship between length of illness and ccf-mtDNA, after controlling for covariates listed above, suggesting that length of illness may be partly responsible for the lower MHI and higher ccf-mtDNA levels observed in BD patients (Table 3). A similar set of analyses, controlling for age, sex, BMI and smoking status, was then performed in the combined BD and nonpsychiatry controls samples, to explore the association of MHI and ccf-mtDNA with functional status. Our results showed that participants with lower MHI and higher ccf-mtDNA levels had worse functional status, highlighting that altered mitochondrial functioning can be signi cant functional de cits (Table 3). Associations from selected models are highlighted in Fig. 3.

Discussion
Recent studies suggest a non-energetic role of mitochondrial as a new paradigm where mitochondria play a role in the adaptation to stress, having an impact on cellular resilience and acting as a source of systemic allostatic load, when the mitochondria lose the ability to recalibrate and maintain cell homeostasis through changes to mitochondrial morphology, dynamics, and function, a process known as mitochondrial allostatic load 8, 28, 29 .
Although our preliminary studies found that BD patients have an impairment in the MQC, the downstream effects of these alterations on the MHI and ccf-mtDNA levels are unknown. To the best of our knowledge, this is the rst study to investigate the MHI in patients with BD and the association of a decrease in the MHI with alterations in the MQC system, ccf-mtDNA levels, and clinical outcomes. This preliminary crosssectional study indicates that BD patients have a decrease in mitochondrial enzymes and MHI, re ecting lower mitochondrial bioenergetic capacity, which could result in increased oxidative stress, decreased mitochondrial Ca 2+ buffering capacity, and loss of ATP, all of which are factors that were described in BD 4,30 . Studies have shown that alterations in proteins involved in mitochondrial fusion and ssion lead to altered mitochondrial shape, loss of mtDNA, decreased mitochondrial respiration, increased oxidative stress, and apoptotic cell death 31,32  capacity. In contrast, mitochondrial ssion is associated with a decrease in the bioenergetic e ciency of the cells.
As described above, maintaining a healthy mitochondrial pool is critically regulated by MQC, and inhibiting the mitophagy pathway could lead to a pronounced accumulation of damaged mitochondria and apoptosis 33 . Moreover, studies have suggested that during the apoptotic process, fragments of mtDNA are released to the extracellular space and may behave as damage-associated molecular patterns (DAMPs) 34,35 , leading to a chronic in ammation without microbial infection -termed sterile in ammation -by the activation of various pro-in ammatory signaling pathways 36-39 . Thus, our second hypothesis was that higher ccf-mtDNA levels would be associated with lower MHI, MQC impairment, and apoptosis activation. Supporting this hypothesis, our preliminary ndings demonstrated that patients with BD had higher plasma levels of ccf-mtDNA, which were negatively correlated with Opa-1 and LC3. Moreover, we found a positive correlation between ccf-mtDNA and Fis-1. Finally, we observed that subjects with lower MHI and higher ccf-mtDNA also had higher active caspase-3 levels. Taken together, our results may indicate that BD patients have an increase in mitochondrial fragmentation associated with a decrease in the mitochondrial bioenergetics' capacity, followed by a lack of mitophagy activation, which could lead to accumulation of damaged mitochondria in the cells, shifting the cell signaling to apoptosis activation. Consequently, fragments of the mtDNA are subsequently released into the circulation and trigger in ammation, a consistent feature of BD's pathophysiology 40 .
Increasing evidence indicates that BD is a systemic disorder that affects not only the brain but also peripheral systems, leading to systemic comorbidities that may represent the result of cumulative "wear and tear" from long-term exposure to stress [41][42][43] . As a result, these comorbidities can in uence and further deteriorate brain function, contributing to accelerated aging and neuroprogression [44][45][46] . In this context, in an exploratory analysis, we investigated the correlation of MHI and ccf-mt-DNA with depressive and manic symptoms severity and functional status. Our results showed that depressive and manic symptoms were associated with lower mitochondrial content and bioenergetic capacity and increased levels of ccf-mtDNA. Our study also found an interesting correlation between worse functional status (as shown by lower GAF and higher FAST scores) with lower MHI and higher ccf-mtDNA levels.
Poor functioning is considered a key factor of disability in patients with BD-l 47 . Furthermore, the length of illness has been found to predict MHI and ccf-mtDNA in BD patients, with every 1-unit increase in length of illness resulting in a decreased MHI of 1.949 and increased ccf-mtDNA of 26.42, after controlling for age, sex, and smoking status. Using a machine learning approach, Sartori et al 48 demonstrated that brain volume changes in MRI were predictors of FAST scores in patients with BD and could identify speci c brain areas related to functioning impairment. Notably, studies have shown that a substantial proportion of patients with BD experience unfavorable functioning in community and outpatient samples, suggesting a signi cant degree of morbidity and dysfunction associated with BD, even during remission periods 49,50 . The FAST score was also reported as sensitive to detecting functioning impairment and accurate in distinguishing early from late stages of BD 51 .
Regardless of whether mitochondrial dysfunction plays a primary or secondary role, our ndings, which align with the literature, indicate that patients with BD present multifactorial alterations in mitochondrial biology. Together, these alterations can lead to mitochondrial allostatic load, which, in turn, contributes to disease progression and poor outcomes via multiple mechanisms, such as gene expression and epigenomics, changes in brain structure and function, abnormal stress reactions, in ammation, and cellular aging. Moreover, because mitochondria are found in all organs, mitochondrial allostatic load affects multiple types of cells and organ systems, resulting in the coexistence of various diseases and symptoms.
There were several limitations in our study. First, we cannot rule out the possibility of a type I error in this preliminary study since it is a cross-sectional study with small sample size. Second, it is known that a single analysis of study participants' peripheral blood is not the most appropriate strategy as directionality and causality cannot be established. Third, the present study did not account for several factors that might impact mitochondrial health dynamically, including lifestyle factors, childhood trauma, chronic stress, suicidal behavior, acute mood states, exercise, and dietary habits. Fourth, the use of PBMC cell mixtures, rather than speci c puri ed cell types, could have reduced our ability to detect mitochondrial alterations with greater sensitivity and speci city among particular cell populations 21 .
Fifth, most patients were taking one or more psychotropic medications, introducing the confounding effect of polypharmacy. Last, despite uncovering an association between MHI and ccf-mtDNA levels with several clinical variables indirectly linked to clinical severity and progression, another limitation of our study includes the absence of information on the number of previous episodes and hospitalizations.
Therefore, our results should be seen as exploratory and require replication and validation. Future research efforts to evaluate mitochondrial functions and related aspects of cellular bioenergetics should be investigated with the proper methodology, using a large sample size, prospective and longitudinal study designs, including repeated measures from patients with BD during all affective states, and statistical adjustment for a range of relevant demographical and lifestyle variables are warranted.
While our ndings need replication, they indicate a role for mitochondrial allostatic load in BD, suggesting that in patients with BD, mitochondria represent a potential biological intersection point that could contribute to impaired cellular resilience, increasing the vulnerability to stress and mood episodes, potentially linking mitochondrial dysfunction with the progression of the disease and poor outcomes.

Declarations
Con ict of interest presented as mean ± standard error of the mean and were analyzed with univariate generalized linear models with adjustment for sex, age, BMI, and smoking status. ** Different from the control group, p < 0.001. Figure 2